Survival, Movement, and Health of Hatchery-Raised Juvenile Lost

Download PDF
0 downloads 0 Views 4MB Size Report
Comment
were homogenized with 1 mL of DI water per 1 g of fish in a blender or a Biospec M133 ...... affects their survival and fitness, and if essential resources are limited, fish will be unable to increase ..... percent20no. percent201037.pdf.] McMahon ...
Prepared in cooperation with the Bureau of Reclamation

Survival, Movement, and Health of Hatchery-Raised Juvenile Lost River Suckers within a Mesocosm in Upper Klamath Lake, Oregon

Open-File Report 2016–1012

U.S. Department of the Interior U.S. Geological Survey

Survival, Movement, and Health of Hatchery-Raised Juvenile Lost River Suckers within a Mesocosm in Upper Klamath Lake, Oregon By Danielle M. Hereford, Summer M. Burdick, Diane G. Elliott, Amari Dolan-Caret, Carla M. Conway, and Alta C. Harris

Prepared in cooperation with the Bureau of Reclamation

Open-File Report 2016–1012

U.S. Department of the Interior U.S. Geological Survey

U.S. Department of the Interior SALLY JEWELL, Secretary U.S. Geological Survey Suzette M. Kimball, Director U.S. Geological Survey, Reston, Virginia: 2016

For more information on the USGS—the Federal source for science about the Earth, its natural and living resources, natural hazards, and the environment—visit http://www.usgs.gov/ or call 1–888–ASK–USGS (1–888–275–8747).

For an overview of USGS information products, including maps, imagery, and publications, visit http://www.usgs.gov/pubprod/. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Although this information product, for the most part, is in the public domain, it also may contain copyrighted materials as noted in the text. Permission to reproduce copyrighted items must be secured from the copyright owner.

Suggested citation: Hereford, D.M., Burdick, S.M., Elliott, D.G., Dolan-Caret, Amari, Conway, C.M., and Harris, A.C., 2016, Survival, movement, and health of hatchery-raised juvenile Lost River suckers within a mesocosm in Upper Klamath Lake, Oregon: U.S. Geological Survey Open-File Report 2016–1012, 48 p., http://dx.doi.org/10.3133/ofr20161012. ISSN 2331-1258 (online)

Contents Abstract ...................................................................................................................................................................... 1 Introduction ................................................................................................................................................................. 2 Methods ...................................................................................................................................................................... 5 Mesocosm Design, Fish Introduction, and Sampling the Mesocosm ...................................................................... 5 Water Quality .......................................................................................................................................................... 8 Movement ............................................................................................................................................................... 9 Mortality and Survival ............................................................................................................................................ 10 Growth and Condition ........................................................................................................................................... 13 Sucker Health and Determination of Cause of Death............................................................................................ 13 Tissue Preparation for Histopathology .................................................................................................................. 14 Results...................................................................................................................................................................... 14 Seasonal Variation in Depth, Temperature, Dissolved Oxygen, and pH ............................................................... 14 Microcystin ............................................................................................................................................................ 18 Un-Ionized Ammonia............................................................................................................................................. 19 Movement ............................................................................................................................................................. 20 Mortality ................................................................................................................................................................ 23 Survival ................................................................................................................................................................. 25 Growth and Condition ........................................................................................................................................... 28 Necropsies ............................................................................................................................................................ 33 Histopathological Evaluation ................................................................................................................................. 35 Discussion ................................................................................................................................................................ 39 Acknowledgments..................................................................................................................................................... 44 References Cited ...................................................................................................................................................... 45

Figures Figure 1. Map depicting location of the mesocosm in Upper Klamath Lake, Oregon ............................................... 6 Figure 2. Water depth and depth of water quality readings in the mesocosm in Upper Klamath Lake, Oregon..... 15 Figure 3. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily water temperatures in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014............................................................... 15 Figure 4. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily dissolved-oxygen concentrations in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014............................. 17 Figure 5. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily pH in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014..................................................................................... 17 Figure 6. Concentrated particulate (a) and dissolved fraction (b) microcystin concentrations approximately 100 meters from the mesocosm in Upper Klamath Lake, Oregon, July 16–September 15, 2014............................ 18 Figure 7. Un-ionized ammonia concentrations approximately 100 meters from the mesocosm in Upper Klamath Lake, Oregon, July 16–September 10, 2014 ........................................................................................................... 19 Figure 8. Diurnal vertical movement patterns of juvenile Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, summarized as percent reads per hour among Surface and Benthos antennas during (a) early (July 2–28), (b) middle (July 29–August 21), and (c) late (August 22–September 15) season ....... 21 Figure 9. Percentage of hourly activity (reads) near benthic and surface antennas of juvenile Lost River suckers and dissolved-oxygen concentration (DO; mg/L) near surface and near benthos in the mesocosm in Upper Klamath Lake, Oregon, on (a) July 14, 2014, and (b) July 15, 2014 ............................................................. 22

iii

Figure 10. Total number of natural daily mortalities for age-1 (groups 1–7) and age-2 (groups 8 and 9) juvenile Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014....................... 24 Figure 11. Total number of Lost River suckers alive in the mesocosm in Upper Klamath Lake, Oregon, July 2– September 15, 2014. Observed reductions in groups are due to natural mortality and scheduled sacrifices .......... 24 Figure 12. Daily survival estimates and 95-percent confidence intervals for age-1 juvenile Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014........................................ 27 Figure 13. Change in standard length and total number of days in the mesocosm for juvenile Lost River suckers in Upper Klamath Lake, Oregon ................................................................................................................. 29 Figure 14. Change in weight and total number of days in the mesocosm for juvenile Lost River suckers in Upper Klamath Lake, Oregon ................................................................................................................ 30 Figure 15. Change in body condition (K = (wt/SL3)*105) and total number of days in the mesocosm for juvenile Lost River suckers in Upper Klamath Lake, Oregon ................................................................................... 31 Figure 16. Whole fish triglyceride levels (mg/g of body tissue) of juvenile Lost River suckers sacrificed directly from the research facility, and at 3 (7-28-14), 6 (8-18-14), and 8 (9-2-14) weeks post introduction from group 2, Upper Klamath Lake, Oregon .................................................................................................................... 32 Figure 17. Gill lamellae from moribund Lost River suckers sampled from the mesocosm in Upper Klamath Lake ................................................................................................................................................ 36 Figure 18. Ichthyobodo sp. parasites (arrows) attached to the gill lamellar epithelium .......................................... 37

Tables Table 1. Group information and size of 395 hatchery-raised Lost River suckers when introduced to the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014................................................................. 7 Table 2. Fates for 395 hatchery-raised Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014 ........................................................................................................................ 7 Table 3. Description of a priori hypotheses tested to describe juvenile Lost River sucker mortality within the mesocosm in Upper Klamath Lake, Oregon ............................................................................................................ 12 Table 4. Average ± standard deviation and maximum number of days one or more fish was alive in the mesocosm, and the total number of days available (from date of introduction to September 15) for each group of hatchery-raised Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014. .................................................................................................................................... 25 Table 5. Model selection results for Kaplan Meier models fit to estimate daily survival of age-1 juvenile Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014....................... 26 Table 6. Prevalence of parasites, bacterial disease, and other abnormalities for moribund (n = 14) and sacrificed, non-moribund (n = 192) Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon............. 33 Table 7. Condition of tissues from necropsies performed in the field immediately following sacrifice for moribund (n = 14) and sacrificed, non-moribund (n = 78) Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon. ............................................................................................................................................ 34 Table 8. Prevalence and percent occurrence of inflammation, necrosis, other histopathological changes, and parasites in gill tissue of Lost River suckers from the mesocosm in Upper Klamath Lake, Oregon .................. 38 Table 9. Prevalence and percent occurrence of inflammation and necrosis in tissues of Lost River suckers from the mesocosm in Upper Klamath Lake, Oregon .............................................................................................. 38 Table 10. Hepatocellular glycogen distribution evaluated by periodic acid-Schiff (PAS) and PAS-diastase staining in Lost River suckers from the mesocosm in Upper Klamath Lake, Oregon. .............................................. 38

iv

Conversion Factors International System of Units to Inch/Pound Multiply

By

To obtain

Length nanometer (nm)

3.93701 × 10-8

inch (in.)

micrometer (µm)

0.000039

inch (in.)

centimeter (cm)

0.3937

inch (in.)

millimeter (mm)

0.03937

inch (in.)

meter (m)

3.281

foot (ft)

kilometer (km)

0.6214

mile (mi)

Area square centimeter (cm2)

0.001076

square foot (ft2)

square centimeter (cm2)

0.1550

square inch (ft2)

Volume microliter (µL)

3.381 × 10-5

ounce, fluid (fl. oz)

milliliter (mL)

0.0338

Ounce, fluid (fl. oz)

liter (L)

33.81402

ounce, fluid (fl. oz)

liter (L)

61.02

cubic inch (in3)

Mass gram (g)

0.03527

kilogram (kg)

2.205

milligram microgram nanogram (ng)

ounce, avoirdupois (oz) pound avoirdupois (lb)

3.5274× 10

-5

ounce, avoirdupois (oz)

3.5274× 10

-8

ounce, avoirdupois (oz)

3.53 × 10

-11

ounce, avoirdupois (oz)

Temperature in degrees Celsius (°C) may be converted to degrees Fahrenheit (°F) as: °F = (1.8 × °C) + 32.

v

Datum Horizontal coordinate information is referenced to the North American Datum of 1983 (NAD 83).

Supplemental Information Concentrations of chemical constituents in water are given in either milligrams per liter (mg/L) or micrograms per liter (µg/L).

Abbreviations AICc

Akaike’s information criterion for small sample size

CV

coefficient of variation

DI

deionized water

DO

dissolved-oxygen

ELISA

enzyme-linked immunosorbent assay

K

body condition K= (wt/SL3)x105

MS-222

tricaine mesylate

NH3

un-ionized ammonia

PAS

periodic acid-Schiff

PIT

passive integrated transponder

PVC

poly(vinyl chloride)

SD

standard deviation

SL

standard length

USGS

U.S. Geological Survey

wt

weight in grams

xg

gravitational force

φ

survival

vi

Survival, Movement, and Health of Hatchery-Raised Juvenile Lost River Suckers within a Mesocosm in Upper Klamath Lake, Oregon By Danielle M. Hereford, Summer M. Burdick, Diane G. Elliott, Amari Dolan-Caret, Carla M. Conway, and Alta C. Harris

Abstract The recovery of endangered Lost River suckers (Deltistes luxatus) in Upper Klamath Lake is limited by poor juvenile survival and failure to recruit into the adult population. Poor water quality, degradation of rearing habitat, and toxic levels of microcystin are hypothesized to contribute to low juvenile survival. Studies of wild juvenile suckers are limited in that capture rates are low and compromised individuals are rarely captured in passive nets. The goal of this study was to assess the use of a mesocosm for learning about juvenile survival, movement, and health. Hatchery-raised juvenile Lost River suckers were PIT (passive integrated transponder) tagged and monitored by three vertically stratified antennas. Fish locations within the mesocosm were recorded at least every 30 minutes and were assessed in relation to vertically stratified water-quality conditions. Vertical movement patterns were analyzed to identify the timing of mortality for each fish. Most mortality occurred from July 28 to August 16, 2014. Juvenile suckers spent daylight hours near the benthos and moved throughout the entire water column during dark hours. Diel movements were not in response to dissolved-oxygen concentrations, temperature, or pH. Furthermore, low dissolved-oxygen concentrations, high temperatures, high pH, high un-ionized ammonia, or high microcystin levels did not directly cause mortality, although indirect effects may have occurred. However, water-quality conditions known to be lethal to juvenile Lost River suckers did not occur during the study period. Histological assessment revealed severe gill hyperplasia and Ichthyobodo sp. infestations in most moribund fish. For these fish, Ichthyobodo sp. was likely the cause of mortality, although it is unclear if this parasite originated in the rearing facility because fish were not screened for this parasite prior to introduction. This study has demonstrated that we can effectively use a mesocosm equipped with antennas to learn about the timing of mortality, movement, and health of PIT-tagged hatchery-raised juvenile Lost River suckers.

1

Introduction Lost River sucker (Deltistes luxatus) was listed as endangered following substantial declines in abundance that occurred over the last century. Historically, these long-lived large-bodied catostomids thrived throughout the Upper Klamath Basin in southern Oregon and northern California (National Research Council, 2004). Suckers were an important fish for Native Americans, especially in Upper Klamath Lake, the largest lentic sucker ecosystem in the basin (Markle and Cooperman, 2002). Population declines were first observed in the mid-1960s when annual recreational harvests were approximately 10,000 fish; a reduction from historical subsistence harvests (Markle and Cooperman, 2002). The extent of the sucker decline was realized 2 decades later when a fish die-off in Upper Klamath Lake revealed a limited age class distribution, which suggested that substantial recruitment into the adult population had not occurred in over a decade (Markle and Cooperman, 2002; National Research Council, 2004). The sucker fishery was closed in 1987, and Lost River suckers were listed as endangered in 1988 (U.S. Fish and Wildlife Service, 1988). Poor water quality, reduced habitat, high nutrient levels, increased abundance of non-native fish species, and changes in algal bloom dynamics are possible causes of population decline. For the largest remaining population of Lost River suckers, recruitment appears to be limited by survival during the juvenile life stage. Lost River suckers are more abundant in Upper Klamath Lake than anywhere else in their range. Recruitment has been closely monitored by trapping, tagging, and tracking the survival of more than 40,000 adult suckers in spawning aggregates from 1999 to 2012 (Hewitt and others, 2014). Fork length data and recruitment estimates from mark-recapture data suggest recruitment into the adult population has been limited during this time (Hewitt and others, 2014). Adult suckers are becoming larger, and previously unseen fish are not smaller, younger recruits. Spawning aggregations of adult Lost River suckers are found in shoreline springs and tributaries to Upper Klamath Lake in March, April, and May (Hewitt and others, 2014). Larvae emerge from the gravels, and those spawned in the river passively drift downstream to Upper Klamath Lake between mid-April and late-May (Ellsworth and Martin, 2012; Hewitt and others, 2014). Age-0 suckers are captured throughout the lake and are relatively abundant until mid- to late September when catches decline (Ellsworth and others, 2009; Bottcher and Burdick, 2010; Burdick and Hewitt, 2012). Very few age-1 or age-2 suckers are captured throughout Upper Klamath Lake in April and June, after which captures of these age classes are extremely rare (Bottcher and Burdick, 2010; Burdick and VanderKooi, 2010; Burdick and Hewitt, 2012). Despite extensive sampling throughout Upper Klamath Lake, abrupt declines in age-0 catches and the scarcity of all other juvenile captures indicate near complete group failure during the juvenile life stage (Bottcher and Burdick, 2010; Burdick and VanderKooi, 2010; Burdick and Hewitt, 2012).

2

Many physical and biological changes have occurred in Upper Klamath Basin during the first half of the 20th century that coincide with the decrease in sucker abundance and these alterations may provide clues to the cause of the species’ decline. Phosphorus levels are naturally high due to the volcanic geology of the region but are augmented by run off from historical and current agriculture and timber harvest practices, which has modified Upper Klamath Lake to a hypereutrophic lake (Eilers and others, 2001; Bradbury and others, 2004; National Research Council, 2004). Sediment cores have revealed changes in Upper Klamath Lake water quality over the last 150 years including changes in nutrients and minerals (Pb, C, N, 15N, P, C, Ti, and Al), diatoms, cyanobacteria (also called blue-green algae), and green algae (Eilers and others, 2001). Large increases in titanium, aluminum, and sediment accumulation during the 20th century are consistent with increased erosional input caused by historical land-use activities (Eilers and others, 2001). Nutrient concentrations of carbon, nitrogen, and phosphorus also have increased over the last 150 years, and diatom communities have exhibited modest changes in species composition (Eilers and others, 2001). One previously abundant taxa of green alga Pediastrum sharply declined while a cyanobacteria Aphanizomenon flos-aquae previously not present in Upper Klamath Lake, has become very abundant (Eilers and others, 2001). Sediment cores indicate that, in more recent history (30–40 years), major changes in water quality and phytoplankton assemblage have occurred (Eilers and others, 2001). Current water-quality conditions in Upper Klamath Lake are driven by large bloom and crash cycles of A. flos-aquae, a cyanobacterium capable of fixing its own nitrogen (National Research Council, 2004; Eldridge and others, 2013). Throughout the summer, A. flos-aquae increases in abundance, relatively uninhibited by nutrient limitations. A massive A. flos-aquae die-off typically occurs following extended periods of calm weather conditions (National Research Council, 2004). Low dissolved-oxygen (DO) concentrations, high temperatures, high pH, and elevated un-ionized ammonia concentrations are potentially stressful water-quality conditions that coincide with large A. flos-aquae blooms and subsequent die-offs (Kann and Smith, 1999; Eldridge and others, 2013). The confounded effect of some or all of these water-quality parameters may contribute to juvenile sucker mortality. Other species of cyanobacteria are present in Upper Klamath Lake to a lesser degree, including Microcystis aeruginosa, a species capable of producing the toxic secondary metabolite microcystin (Eldridge and others, 2013). M. aeruginosa is usually less than 1 percent of the total phytoplankton biomass but concentrations increase after the A. flos-aquae die-off and nitrogen becomes available through decomposition (Eldridge and others, 2013). Suckers in Upper Klamath Lake could be exposed to microcystin by swimming in contaminated water or by consuming the toxinogenic cells. Microcystin toxins target the liver but laboratory and field studies of rainbow trout (Oncorhynchus mykiss), whitefish (Coregonus lavaretus), zebrafish (Danio rerio), and other fish species have shown that microcystins can also affect other organs, growth, behavior, osmoregulation, and heart rate at different life stages (Malbrouck and Kestemont, 2006). Toxins from M. aeruginosa have been linked to tumor promotion and various diseases in wildlife and domestic animals (Carmichael, 1994; Falconer, 1999). For juvenile and adult life stages of fish, immersion in microcystin-contaminated water is less harmful than consuming affected cells (Malbrouck and Kestemont, 2006). Lost River suckers are benthic feeders and could be consuming M. aeruginosa from the substrate (Eldridge and others, 2013). It is unclear if microcystin toxins are contributing to high juvenile mortality in Upper Klamath Lake as preliminary laboratory feeding trials have produced inconclusive results (Barbara A. Martin, U.S. Geological Survey, oral commun., 2015).

3

Juvenile sucker survival is difficult to estimate and factors affecting survival are difficult to study in this large ecosystem. Understanding survival for juveniles in Upper Klamath Lake is especially difficult because captures are low, distribution is patchy, and captures decline rapidly throughout the season. Mark-recapture methods of survival estimation are limited by sample size and re-detection rates. In general, fewer than 250 juveniles large enough [≥ 70 mm standard length (SL)] to tag with a passive integrated transponder (PIT) tags are captured each year, and redetection rates for these juveniles are low. For example, one study that tagged 592 juvenile suckers in Upper Klamath Lake from 2009 to 2012 redetected only 6.7 percent of tags on remote PIT tag antennas and on scans of bird colonies (Burdick, 2013). Additionally, many of the factors affecting juvenile survival and health such as parasites, disease, and poor water quality could be causing the biggest challenge associated with studying survival, low capture rates. An alternate method to study juvenile health and survival is to introduce juvenile Lost River suckers into a mesocosm within the natural environment. Typically, mesocosms isolate part of a natural ecosystem and allow researchers to control or manipulate some aspects of the study environment (Horne and Goldman, 1994). Individual fish can be closely monitored in mesocosms, and sample sizes are easily manipulated. Many studies have used mesocosms to monitor survival, growth, movement, and understand interspecific relationships (Welker and others, 1994; Williams and others, 2002). In this study, we used one large mesocosm to monitor the health, daily survival, and hourly movement of hatchery-raised Lost River suckers in Upper Klamath Lake. The specific objectives for this pilot study were to (1) assess our ability to quantify temporal survival for hatchery-reared juvenile Lost River suckers within a mesocosm in Upper Klamath Lake, (2) assess our ability to identify changes in health, growth, and body condition of suckers throughout the summer, (3) assess our ability to study movement patterns in relation to water-quality conditions, and (4) to summarize how survival, health, and movement patterns varied throughout the summer and relative to the water-quality parameters we measured. To estimate survival of age-1 suckers, we monitored daily movement and survival of PITtagged fish with vertically stratified remote sensing antenna arrays in a mesocosm within Upper Klamath Lake. Healthy and moribund individuals were sacrificed to assess general health from July to mid-September 2014. Water-quality parameters were measured near the surface (30 cm from the top of the water column) and near the benthos (30 cm from the bottom of the water column) throughout this study to assess how sucker survival and behavior was affected by changes in water quality. Un-ionized ammonia and microcystin concentrations were quantified weekly to monitor seasonal trends in relation to sucker survival, health, and behavior.

4

Methods Mesocosm Design, Fish Introduction, and Sampling the Mesocosm We deployed a 3 × 3 × 3 m net pen (hereinafter “mesocosm”) in the Fish Banks area of Upper Klamath Lake (fig. 1). Several springs enter Upper Klamath Lake in this area, so water quality is relatively good. We chose to place the mesocosm at Fish Banks because fish would likely survive at least part of the study duration in this location. The frame of the mesocosm was constructed of 7.62 cm PVC. A large net made of 0.64 cm2 mesh (20 kg) nylon was manufactured with corner-to-corner sleeves that held the PVC frame. A buoyant dock 1- to 2-m wide surrounded the mesocosm, kept the mesocosm anchored in one location, and provided researchers easy access to the mesocosm. Bird netting with 2.5 cm2 mesh was hung across the top of the mesocosm to prevent avian predation. Three vertically stacked antennas (1.2 × 0.6 m) detected PIT-tagged suckers within the center of the mesocosm. These antennas were placed (1) near the water surface, (2) in the middle of the water column, and (3) on the benthos. Additionally, two smaller (0.3 × 0.3 m) antennas were placed in opposite corners on the bottom of the mesocosm to track horizontal benthic movement. To avoid overloading data storage capabilities, once a tag was detected on an antenna, subsequent detections of the same tag on the same antenna were not recorded for 30 minutes. A detection of the same tag on another antenna would restart the clock on all antennas, such that rapid movement was still detected. Lost River suckers propagated from wild adult Lost River suckers in 2012 and 2013 were raised from eggs at the Coleman National Fish Hatchery (U.S. Fish and Wildlife Service, Anderson, California) and kept as juveniles at the Klamath Tribes Fish Research Facility (Chiloquin, Oregon). We sedated 301 age-1 and 120 age-2 juvenile suckers with tricaine mesylate (MS-222) and tagged all fish with 12-mm, 134.2 kHz PIT tags. Biomark® MK25 implant guns were used to insert preloaded needles into the body cavity between the pectoral fins on the underside of the fish. All tags were injected from the anterior end towards the posterior end to keep the needle and tag away from the heart. All fish were held in two research tanks (one for each age class) to recover from tagging for at least 20 days. Within the first week, 16 (less than 4 percent) fish died from tagging, and 5 fish shed their tags. No additional mortalities occurred post tagging after the first week in the research tanks. Tag failure was rare (n = 1) but, when it occurred, the fish was removed from all analyses due to low detection rates. Release batches (hereinafter “groups”) of juvenile Lost River suckers were introduced to the mesocosm weekly from July 2 to September 2 (table 1). Staggered entry was implemented to test the carrying capacity of the mesocosm and to avoid any unexpected pitfalls associated with introducing all fish at once (for example, early season mortality). All fish were scanned for a PIT tag, measured to SL, and weighed prior to entry. Fish ranged in SL from 90 to 177 mm and in weight from 7.6 to 72.9 g (table 1). Each week, a subset of non-moribund fish (n = 5–15) were removed from the mesocosm and sacrificed to assess changes in health or triglycerides. The mesocosm was sampled by lifting two or three bottom corners onto the dock and dip-netting suckers out of the mesocosm net pen. During weekly sampling, all netted fish were scanned and either returned to the mesocosm or sacrificed (table 2). At the end of the study (September 15), all fish still alive were sacrificed.

5

Figure 1. Map depicting location of the mesocosm in Upper Klamath Lake, Oregon.

6

Table 1. Group information and size of 395 hatchery-raised Lost River suckers when introduced to the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014. [SL, standard length; wt, weight] Group

Group size Date introduced Age1 SL2,3 wt3,4

1

2

3

4

5

6

7

8

9

All

20 7-02-14 1 97.5 ±4.8 13.0 ±2.0

45 7-07-14 1 100.8 ±4.4 13.7 ±2.2

45 7-14-14 1 100.4±4.6 13.7 ±2.3

50 7-21-14 1 100.6 ±5.5 14.1 ±2.4

54 7-28-14 1 102.3 ±5.3 14.8 ±2.4

55 8-04-14 1 104.4 ±5.8 15.6 ±2.9

8 8-11-14 1 104 ±5.3 15.9 ±2.8

55 8-25-14 2 161.3 ±7.0 51.3 ±8.2

63 9-02-14 2 156.1 ±7.2 49.9 ±8.6

395 1–2 118.6 ±26.8 25.2 ±17.3

1

Years.

2

Standard length, in millimeters.

3

Average ±standard deviation.

4

Weight in grams.

Table 2. Fates for 395 hatchery-raised Lost River suckers in the mesocosm in Upper Klamath Lake, Oregon, July 2–September 15, 2014. Fate Found dead1,2 Moribund fish2,3 Sacrificed for histology2 Sacrificed for triglycerides2 Handling mortality2 Accidently released2 Defective tag4 Sacrificed at end of study5

1 9 2 0 0 0 1 0 8

2 17 0 16 12 0 0 0 0

3 24 2 6 0 0 0 0 13

4 40 3 6 0 1 0 0 0

Group 5 6 35 51 3 0 9 3 0 0 5 0 0 0 0 1 2 0

7 4 0 4 0 0 0 0 0

8 6 2 8 0 0 1 0 38

1

Natural mortality.

2

Throughout study (excludes sacrifices on September 15, 2014).

3

Preserved for histology.

4

The individual fish with a defective tag was not included in the survival or movement analyses.

5

A subset of these fish were preserved for histology.

7

9 8 2 4 0 0 0 0 49

All 194 14 56 12 6 2 1 110

Water Quality Water-quality variables were measured hourly near the top (surface) and bottom (benthos) of the water column in order to examine sucker response to changing conditions. Temperature, DO concentration, and pH were measured hourly from July 9 to September 15, 2014, on four YSI 600XLMM sondes. Sondes were positioned 30 cm from the top (hereinafter, “Surface”) and bottom (hereinafter, “Benthos”) of the water column. Sonde depths were adjusted each Wednesday, when necessary, to account for declining lake elevations and mesocosm depth. Field sondes were retrieved each week, cleaned, and exchanged with laboratory-calibrated replacement field sondes. Upon retrieval, a reference sonde was used to measure water-quality parameters at Surface and Benthos sonde sites. Field sondes were cleaned and water-quality parameters of field and reference sondes were recorded to quantify deviance. When field retrieval readings deviated substantially (temperature > 0.4 °C, DO > 0.8 mg/L, or pH > 0.4) from reference sonde readings, minor adjustments in that water-quality parameter field readings were made by regressing the difference across the week of readings. Instances of substantial deviation were rare, and small deviations were not altered. Spurious outlier data points were removed. To assess sucker health and survival relative to un-ionized ammonia concentrations, we collected weekly point water samples in the middle of the water column using a peristaltic water pump, calibrated sonde, and acid washed Masterflex® tubing. All field and laboratory equipment used to sample un-ionized ammonia and microcystin was cleaned in the laboratory using LiquinoxTM (Alconox) according to U.S. Geological Survey (USGS) protocol (U.S. Geological Survey, 2004). Ammonia tubing also was acid washed with a 5 percent HCl solution in the laboratory then rinsed with 900 mL of lake water prior to sample collection. Sample bottles were rinsed with deionized (DI) water three times and filled with approximately 60 mL DI water for transport to the field. Capsule filters were flushed with 2 L of DI water in the laboratory. Samples (125 mL) were collected after sample bottles were rinsed with three field rinses through one single capsule filter. Replicate samples were collected using a second capsule filter. Replicate and split samples were collected alternate weeks for quality control. Blank un-ionized ammonia samples were collected each week prior to the collection of field samples using the same protocol as field samples and inorganic blank water (Ricca Chemical Company LLC). Un-ionized ammonia samples were stored on ice during transport and refrigerated at 4 °C for up to 2 days until accepted by the Klamath Tribes Sprague River Water Quality Laboratory. Samples were then refrigerated for up to 10 days at 4 °C before they were analyzed. Discrete analyzer methods by colorimetric determination were used to quantify un-ionized ammonia as nitrogen by the Sprague River Water Quality Laboratory according to U.S. Environmental Protection Agency (1979) protocols. (Craig Spoonemore, Klamath Tribes Sprague River Water Quality Laboratory, oral commun., 2015). The more toxic un-ionized ammonia (NH3) concentrations were calculated from total un-ionized ammonia concentrations using pH and water temperature measured during sample collection (Fairchild and others, 2005).

8

To assess sucker health and survival relative to microcystin concentrations, we collected 1 L (two 500 mL samples) of depth-integrated water samples each week. Split samples were collected from the same sample as the primary to test for within sample variation while replicate samples were collected from a second water sample to test for environmental variation. Blank samples were collected on July 16, August 13, and September 10, 2014, prior to the collection of any field samples using organic blank water (EMD Millipore Corporation©). All sample collection equipment, including cage sampler bottles, churn splitter, and 500 mL borosilicate amber glass sample bottles, were field-rinsed prior to sample collection. All samples were depth-integrated and churned according to USGS protocol during collection (U.S. Geological Survey, 2006). Field microcystin samples were filtered through a 63-μm sieve to separate colonies of phytoplankton from the rest of the sample. The fraction of the sample retained on the sieve was considered the particulate fraction and the filtrate from the sieving process was considered the dissolved fraction. The particulates were concentrated and re-suspended in tap water after filtering. The total volume of the sample and the volume of tap water used to concentrate and re-suspend the particulates were recorded and later used to calculate the concentration factor. Both particulate and dissolved fractions were processed through three freeze-thaw cycles to lyse cells prior to analysis. Samples were filtered through either a 0.45- or 0.30-μm filter, depending on the filtering supplies available at the time. Filters were either 0.45-μm UniPrepTM syringeless glass microfiber filters (Whatman, Inc., Clifton, New Jersey) or 0.30-μm pre-fired glass filters (Advantec®). The 0.30-μm filters were used with a glass filter holder, glass filter flask, and vacuum hand-pump. Dilutions of both particulate and dissolved fractions were performed, when necessary, to bring the sample within the detection range of the analysis. Particulate and dissolved fractions were both analyzed for microcystin concentration using an enzymelinked immunosorbent assay (ELISA, kit PN 520011, Abraxis® LLC, Warminster, Pennsylvania). Sample absorbances were measured at 450 nm and calibration standards were analyzed with the samples. Microcystin concentrations were calculated using the regression from the calibration standards. Concentrated particulate results were multiplied by the concentration factor (volume of re-suspended particulates/volume of field sample) to determine the final microcystin concentration for the particulate samples.

Movement Daily movement patterns of all living suckers were summarized by summing the number of remote contacts within each hour at the benthos and surface antennas and divided by the total number of live remote contacts from these antennas to obtain an activity index at each depth. These diurnal movement patterns were separated into three subseasons—Early (July 2– 28), Middle (July 29– August 21), and Late (August 22 – September 15) to assess variation in diurnal movement throughout the season.

9

Mortality and Survival Movement patterns of found dead fish were individually assessed and each fish was assigned an expected date and time of mortality based on individual movement patterns. In all cases, living remote detection patterns exhibited frequent (≤ 30 minutes) detection among antennas, whereas non-living remote detection patterns exhibited intermittent to frequent detection at one antenna for prolonged (hours to days) periods. In most cases (96 percent), mortality was confirmed within 5 days of when the body was retrieved from the mesocosm. The design of this study was not intended to assess mechanisms causing variation in mortality among groups nor to assess how mortality varied by age (groups 1–7 versus 8–9). The purpose of introducing fish in groups was to estimate the carrying capacity of the mesocosm for future studies and to assess the ability of the technology to manage differential loads. Kaplan-Meier (known fate) models were used to estimate daily survival of 276 age-1 Lost River suckers in the mesocosm over 76 days (Kaplan and Meier, 1958). Remote detections among antennas were used to create live-dead encounter histories and competing models were ranked in Program MARK. Pollock’s staggered entry design was used to specify temporal variation in groups. Survival parameters for these groups were fixed (ϕ = 1) prior to introduction of each group. Healthy sacrificed individuals were censored out of the population and did not contribute to subsequent survival estimates. Moribund sacrificed individuals were not censored out of the population and were assigned a mortality date as the date collected. The assumptions associated with staggered entry Kaplan-Meier models of tagged animals are: 1. tags are not lost, 2. tag failure does not occur, 3. tagged population is a random and a representative sample of the larger population, 4. detection probability is 1.0, 5. status (alive or dead) of each individual is known at each sampling occasion (each day for this study), 6. time intervals are similar among occasions, 7. individual animals have independent survivals, 8. newly tagged or introduced animals have the same probability of surviving to time (t+i) as animals already present in the study, 9. survival rates are constant over defined intervals (1 day in this study), 10. survival does not vary among groups in staggered entry, and 11. censoring is random and not related to fate.

10

The design of this study made assessing possible violations easy. We had no instances of tag loss within the study because we allowed tag wounds to heal 20 or more days prior to introduction, and all fish were screened for tag presence prior to being introduced to the mesocosm (1 and 8). We had one instance of tag failure where the tag was cracked and the fish was only detected four times in 7 days (2). This fish was not included in any analyses because its date of mortality and movement patterns could not be assessed. The population was fully represented because the tagged population was the entire population (3). We were able to account for all but one (cracked tag) age-1 fish (276/277) each day (4). The frequent detections among antennas allowed us to easily identify the day (sampling occasion) of mortality (5). All time intervals were short (24 hours) and representative of our ability to estimate timing of mortality (6 and 9). We assumed independence of survival among individuals based on the biology of the species because we did not directly observe schooling behavior (7). We accounted for variation in survival among newly introduced fish by testing for a group effect in the model design (8 and 10). Censoring was random because up to five fish were randomly selected for sacrifice 2 and 4 weeks after introduction (11). Fish with external lesions or parasites were not preferentially sacrificed unless they were moribund in which case they were not censored (11). A set of a priori water quality hypotheses were tested against full and simplified group and time models to describe potential causes of daily sucker mortality. We tested daily, weekly, linear, and quadratic time trends. Daily water quality hypotheses included conditions such as mean, minimum, or maximum DO concentration, temperature, or pH, and weekly microcystin or un-ionized ammonia concentrations (table 3). In addition, we tested three hypotheses summarizing total hours each day above 25 °C, or pH 9, or DO concentrations below 4.0 mg/L. These values are extreme relative to the conditions observed in the mesocosm yet conservative in relation to known high stress thresholds for Lost River suckers (28 °C, pH of 9.75, and DO concentrations less than 2.1 mg/L; Loftus, 2001; Meyer and Hansen, 2002). All water-quality parameters were standardized [𝑥̅ = 0.0, standard deviation (SD) = 1.0] and assessed as time varying covariates. We also tested a model that tested for stress induced mortality on sampling days. Models with inestimable parameters were removed from the model set and are not included in the Akaike’s information criterion (AICc) table. All models were run using the logit link function in program MARK. All models were ranked using AICc to account for small sample size (Burnham and Anderson, 2002). Models within two AICc units were considered competitive (Burnham and Anderson, 2002). The relative likelihood of each model was assessed using Akaike weights, which rank the probability of each proposed model in a model set given the data (Burnham and Anderson, 2002). Models with high Akaike weights have more evidence to support the relative likelihood of that model (Burnham and Anderson, 2002). Model pairs can be compared using evidence ratios, which are the ratio of one model’s Akaike weight relative to another (Burnham and Anderson, 2002). Evidence ratios that are low suggest model selection uncertainty while high evidence ratios suggest strong support for one model relative to a second model (Burnham and Anderson, 2002). Classical methods for estimating goodness-of-fit (Program RELEASE, median ĉ, bootstrapping) are not available for known fate models, so we investigated the robustness of model rankings by inflating ĉ from 1.0 (no dispersion) to 3.0 (extreme over dispersion) to simulate different amounts of dispersion as ranked in the quasi-AICc (Devries and others, 2003; Smith and others, 2015).

11

Table 3. Description of a priori hypotheses tested to describe juvenile Lost River sucker mortality within the mesocosm in Upper Klamath Lake, Oregon. [φ, survival] Model type Group and time

Water quality

Null

Model name

Description

Number of parameters

φ (g * t) φ (g + t) φ (g) φ (g * quad) φ (g + quad) φ (g * lin) φ (g + lin) φ (t) φ (week) φ (quad) φ (lin) φ (samp.stress)

interactive group and daily time additive group and daily time group interactive group and quadratic time trend additive group and quadratic time trend interactive group and linear time trend additive group and linear time trend daily time weekly time quadratic time trend linear time trend excess stress induced on sampling days

φ (temp.mean) φ (temp.max) φ (temp.cv) φ (temp.25) φ (DO.mean) φ (DO.min) φ (DO.cv)

2 2 2 2 2 2 2

φ (DO.4) φ (pH.mean) φ (pH.max) φ (pH.cv) φ (pH.9) φ (NH3) φ (conc.micro) φ (diss.micro)

mean daily temperature maximum daily temperature daily temperature coefficient of variation total hours of temperature greater than 25˚C per day mean dissolved-oxygen concentrations minimum dissolved-oxygen concentrations daily dissolved-oxygen concentrations coefficient of variation total hours of dissolved oxygen concentration less than 4 mg/L per day mean daily pH maximum daily pH daily pH coefficient of variation total hours of pH greater than 9 per day weekly un-ionized ammonia concentrations weekly concentrated particulate microcystin concentration (ppb) weekly dissolved fraction microcystin concentration (ppb)

φ (.)

constant model

1

12

532 82 7 15 15 8 8 76 11 3 2 2

2 2 2 2 2 2 2 2

Growth and Condition Individuals found dead were in various states of decay, which resulted in some fish lengths being unattainable. Length and weight measurements were taken from all sacrificed suckers. When possible, standard, fork, and total length measurements were taken from suckers retrieved shortly after death. SL and weight (wt) were used to compute body condition (that is, the Fulton condition factor; K = [wt/SL3]×105) and assess growth. Total changes in SL, weight, and body condition were plotted for individuals that spent more than 3 weeks in the mesocosm, and linear regressions were fit to these data. A total of 12 suckers from group 2 were sacrificed using an overdose of MS-222, 3 (n = 5), 6 (n = 5), and 8 (n = 2) weeks following introduction for whole-body triglyceride analysis. One sample (sacrificed during week 6) was accidentally discarded before it was quantified (triglycerides analyzed, n = 4). Five age-1 suckers from the Klamath Tribes Fish Research Facility, which were never introduced to the mesocosm, also were sacrificed for triglyceride analysis. These fish represented a starting condition for comparison with mesocosm fish. A decrease in triglycerides would indicate a lack of feeding in the mesocosm, whereas a flat or increasing trend in triglycerides over time would indicate adequate food resources in the mesocosm. Suckers were transferred from the field or fish research facility to the laboratory on dry ice where they were stored at -80 °C until they could be processed. Fish were homogenized with 1 mL of DI water per 1 g of fish in a blender or a Biospec M133 homogenizer. A small subsample (0.2–0.5 g) of fish puree was diluted at a 4 to 1 ratio of isopropyl alcohol (2propanol) to fish sample. Samples were rotated for 20 minutes and centrifuged for 5 minutes at 3,220 gravitation force. Diluted supernatant was assayed for triglycerides (milligrams of triglycerides per gram of tissue) using a Powerwave 340 colorimeter (BIO-TEK), Pointe Scientific triglyceride glycerol phosphate oxidase (GPO) kits, and methods from Weber and others (2003). A t-test was used to test for differences in triglyceride levels between mesocosm suckers and wild, age-0 suckers captured near the mesocosm during the summer 2014.

Sucker Health and Determination of Cause of Death Throughout this study, 23.3 percent (n = 92) of suckers were sacrificed for histological examination. Of these, 33 were selected a posteriori for full histopathological examination. Moribund (n = 14) fish collected from July 29 to September 11 and non-moribund (n = 19) fish collected August 4 and August11 during the July 28–August 16 mortality event were identified as high priority and were used to assess fish health in the mesocosm. Fish were sacrificed using an overdose of MS-222, external lesions and parasite loads were noted, and SL was measured. Field necropsies were performed postmortem on fish selected for histological examination to aid laboratory evaluation. Field necropsy evaluations identified gill color and shape, liver color and texture, spleen color, gall bladder color and distention, and presence and amount of visceral fat (Goede and Barton, 1990). Necropsies were quickly performed and all samples were preserved in Carson’s modified Millonig phosphate-buffered formalin (Carson and others, 1973) within 5 minutes post-sacrifice. Fish were stored in formalin for 3–5 days and transferred to 70 percent ethanol. Samples were weighed in the laboratory.

13

Tissue Preparation for Histopathology Tissues collected for histopathological analysis included the gills, heart, anterior kidney, posterior kidney, liver, spleen, pancreatic tissue, and gastrointestinal tract. Skin and skeletal muscle were collected from fish with external lesions. All tissues were subjected to routine paraffin processing, sectioned at 5 µm and stained with Gill’s No. 3 hematoxylin and eosin (ThermoFisher Scientific) according to the manufacturer’s instructions to determine the degree of host response and parasite occurrence. All tissues were examined by light microscopy with a Zeiss Axiophot photomicroscope, and the degree of host response, including inflammation and cell necrosis per 200× field, was recorded and scored using a four-point scale. Tissue distribution of host response was scored as none, focal, multifocal or diffuse; severity was scored as none, minimal to mild, moderate, or severe. The location and identification of parasites and the degree of host response to parasites also was recorded. The presence, appearance of vacuoles and staining characteristics of hepatocyte cytoplasm were used to explore liver energy storage as another possible indicator of the nutritional status of fish. Liver tissue was stained with periodic acid-Schiff (PAS) and PAS-diastase to estimate the amount of glycogen stored in hepatocyte cytoplasm. Glycogen distribution was rated as none, focal (low levels), or multifocal to diffuse (high levels). The presence of hepatocellular vacuoles morphologically consistent with lipid storage also was noted.

Results Seasonal Variation in Depth, Temperature, Dissolved Oxygen, and pH The total depth of the mesocosm decreased by 0.5 m throughout the season (fig. 2) as lake elevation in Upper Klamath Lake declined. As a result, the water-quality readings from the Benthos sonde were taken at decreasing depths as the season progressed (fig. 2). The sharp peak in depth readings from August 18 to 20, 2014, was due to improper replacement of the sonde chain after sacrificing the suckers for the week. During this time, depths were approximately 20 cm below the surface and 40 cm above the benthos. Throughout the entire season, hourly water temperature ranged from 15.01 to 29.86 °C. Mean daily water temperature generally decreased throughout the study season, but ranged from 17.55 to 25.62 °C. Maximum daily water temperature ranged from 19.55 to 29.86 °C. On average, daily temperatures fluctuated 4.7 ±1.1 °C SD in the mesocosm throughout the season (fig. 3). Diel temperature differences between the Surface and Benthos sondes were similar to each other; the mean difference in hourly temperature was 0.38 ±0.61 °C. Temperature was highest July 9–19 when all hourly readings ranged from 21.8 to 29.9 °C (fig. 3). At the end of the study season, in September, daily water temperatures ranged from 15.01 to 22.57 °C. Cool weather conditions on July 23–24 and September 12–14 caused decreases in water temperature (fig. 3). Generally, the diel temperature coefficient of variation (CV) increased slightly throughout the season. The mean diel temperature CV was 0.07 ±0.02 SD and ranged from 0.03 to 0.13. There was little variation in the standard deviation of diel temperature throughout the season (1.46 ±0.43, range = 0.67–2.65).

14

Figure 2. Water depth and depth of water quality readings in the mesocosm in Upper Klamath Lake, Oregon. Water quality samples collected July 9–September 15, 2014.

Figure 3. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily water temperatures in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014.

15

Mean daily DO concentrations (8.97 ±1.74 mg/L) and mean diel fluctuations in DO concentrations (5.81 ±2.37 mg/L) varied throughout the sampling season (fig. 4). Generally, DO concentrations were similar between Surface and Benthos sondes (average difference 0.75 ±1.18 mg/L) except from July 12 to 15 when daily fluctuations were greatest (range = 8.14–14.46 mg/L, minimum for period = 0.07, maximum for period = 16.25 mg/L) and concentrations reached seasonal lows (fig. 4). Extreme differences in DO concentration between the Surface and Benthos during this time may represent brief occurrences of lake stratification. DO concentrations usually increased and decreased gradually throughout the day, but on some days DO concentrations changed rapidly. Rapid increases may represent instances of lake stratification followed by mixing. Mean minimum daily DO concentrations were 6.21 ±2.16 mg/L in the mesocosm (fig. 4). There was little variation in the diel CV in DO concentrations throughout the season (0.19 ±0.10, range = 0.08–0.63), although diel CV was high July 14–15 (0.61–0.63). There was little variation in the SD of diel DO concentrations throughout most of the season (1.65 ±0.72, range = 0.63–4.67), although the SD was relatively high in mid-July. DO concentrations were less than 4.0 mg/L for 15 hours near the Surface and 72 hours near the Benthos over the entire 1,622 hour study period. DO concentrations were less than 4.0 mg/L at both Surface and Benthos sondes for 10 hours throughout the study season. The longest duration of DO concentrations less than 4.0 mg/L was on July 31 between 0400 and 1000 hours, when concentrations ranged from 3.28 to 3.93 mg/L. Mean pH in the mesocosm was high (8.98 ± 0.41) and ranged from 7.32 to 9.75 (fig. 5). Throughout the season, pH was often greater than 9, although pH was relatively low July 27–August 4. On average, pH fluctuated 0.81 ±0.42 (average ± SD) per day, although there were several instances throughout the study (July 23–26, August 7–9, and August 29–September 15) when pH was high and diel fluctuations were small (fig. 5). Mean maximum daily pH was 9.35 ±0.25 and ranged from 8.63 to 9.75. Mean minimum daily pH was 8.54 ± 0.54 and ranged from 7.32 to 9.26. The mean difference between Surface and Benthos pH was 0.17 ±0.19. Generally, pH at the Surface and Benthos fluctuated in tandem, but pH was usually slightly higher at the Surface than at the Benthos.

16

Figure 4. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily dissolved-oxygen concentrations in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014.

Figure 5. Mean (black circle), minimum (bottom bar), and maximum (top bar) daily pH in the mesocosm in Upper Klamath Lake, Oregon, July 9–September 15, 2014.

17

Microcystin Environmental variability of microcystin was low within sampling occasions. The difference between primary and replicate samples of dissolved fraction microcystin was 0.22 ±0.16 μg/L and ranged from less than 0.001 to 0.440 μg/L. The difference between primary and replicate samples of concentrated particulate microcystin was 1.08 ±1.87 μg/L and ranged from 0.04 to 3.9 μg/L. Variability within samples of microcystin also was low. The difference between primary and split samples of dissolved fraction microcystin was 0.07 ±0.07 μg/L and ranged from 0 to 0.17 μg/L. The difference between primary and split samples of concentrated particulate microcystin was 0.34 ±0.36 μg/L and ranged from 0.01 to 0.79 μg/L. Low variation between primary and replicate or primary and split samples suggest quality control was adequate throughout the season. Microcystin concentrations in the particulate fraction (≥ 63 µm) increased earlier and peaked much higher than the dissolved fraction. Concentrated particulate microcystin was low in July, increased quickly in August, and was highest (43.22 μg/L) August 20, 2014 (fig. 6a). Concentrated particulate microcystin levels stayed relatively high until the last week of the study (September 15) when concentrations decreased to less than 5 μg/L. Weekly samples of dissolved fraction microcystin were less than 1 μg/L through the week of August 17. Following 2 weeks of elevated concentrations in the particulate fraction, dissolved fraction concentrations increased sharply to 4.55 μg/L the week of August 20. Dissolved concentrations remained greater than 2 μg/L for 3 weeks in August and September (fig. 6b).

Figure 6. Concentrated particulate (a) and dissolved fraction (b) microcystin concentrations approximately 100 meters from the mesocosm in Upper Klamath Lake, Oregon, July 16–September 15, 2014. Note the difference between y-axis scales in these two graphs.

18

Un-Ionized Ammonia Environmental variability of un-ionized ammonia was low within sampling occasions. The difference between primary and replicate samples of un-ionized ammonia was 1.97 ±1.76 µg/L, and ranged from less than 0.85 to 4.60 µg/L. Variability within samples of un-ionized ammonia also was low. The difference between primary and split samples of un-ionized ammonia was 2.63 ±3.54 µg/L and ranged from 0 to 7.83 µg/L. Low variation between primary and replicate or primary and split samples suggest quality control was adequate throughout the season. Un-ionized ammonia concentrations varied among weeks; concentrations were highest July 16 (103 µg/L), decreased throughout July (39–41 µg/L), and were very low (< 30 µg/L) throughout August and September (fig. 7).

Figure 7. Un-ionized ammonia concentrations approximately 100 meters from the mesocosm in Upper Klamath Lake, Oregon, July 16–September 10, 2014.

19

Movement Suckers in the mesocosm exhibited vertical diurnal movement patterns. Suckers were detected most often on benthos antennas during daylight hours and throughout the water column at night (fig. 8). Suckers settled near the benthos around sunrise or between approximately 0400 and 0600 hours and began using the entire water column again around sunset, or between approximately 2000 and 2100 hours. This behavior was observed throughout the summer, but daytime use of the benthos was most distinct during late summer (August 22–September 15). The timing of movement patterns was not associated with diurnal fluctuations in DO concentrations, pH, or temperature. In the early season (July 2–28), DO concentrations, pH, and temperature generally were higher near the surface than near the benthos, including when suckers were using the benthos and the whole water column. In mid-season (July 29–August 21), DO concentrations were on average higher near the benthos than the surface, while pH and temperature generally were the same throughout the water column when suckers spent more time on the bottom between 2100 and 0400 hours. Finally, late in the season (August 22–September 15), DO concentrations were higher and temperature and pH were lower near the benthos than near the surface when suckers were on the bottom. The most extreme examples of non-avoidance of poor water quality occurred twice when DO concentrations were less than 1.58 mg/L, the median 24-hour lethal concentration (LC50) for juvenile Lost River suckers (Saiki and others, 1999). DO concentrations near the benthos were low on July 14 (0.85 mg/L at 1100 hours) and July 15 (